基于超高压输电线路智能巡检的建筑物提取方法研究  

Study on Building Extraction Method for Intelligent Inspection of Ultra High Voltage Transmission Lines

作　　者：翟春雨 张国君 任雅茹 王大鹏 白翔宇[2] 陈顺华 林恒立 ZHAI Chunyu;ZHANG Guojun;REN Yaru;WANG Dapeng;BAI Xiangyu;CHEN Shunhua;LIN Hengli(Inner Mongolia Ultra High Voltage Power Supply Branch of Inner Mongolia Electric Power(Group)Co.,Ltd.,Hohhot 010000,China;Inner Mongolia University,Hohhot 010000,China;Beijing SensingTerra Technology Co.,Ltd.,Beijing 100085,China)

机构地区：[1]内蒙古电力(集团)有限责任公司内蒙古超高压供电分公司,内蒙古呼和浩特010000 [2]内蒙古大学,内蒙古呼和浩特010000 [3]北京星视域科技有限公司,北京100085

出　　处：《电工技术》2024年第16期209-212,共4页Electric Engineering

基　　金：内蒙古自治区科技计划资助(编号2022YFSJ0004)。

摘　　要：在超高压输电线路智能巡检过程中,提取卫星遥感影像中的建筑物至关重要。针对该任务提出一种细粒度特征融合的建筑物提取方法--U-DeepNet。该方法集成了Deeplab V3+与U-Net模型的优点,充分提取细粒度特征,确保密集排列的建筑物亦能被准确提取与区分开来。通过添加2个不同的残差U型块(RSU),增加了网络对小目标的灵敏感知,并使用不同维度的注意力机制,合理关注有效信息。在马萨诸塞州建筑物数据集上的实验验证了该方法的优越性。In the process of intelligent inspection of ultra-high voltage transmission lines,it is crucial to extract buildings from satellite remote sensing images.In this paper,we propose a fine-grained feature fusion building extraction method,U-DeepNet,which integrates the advantages of Deeplab V3+and U-Net model to fully extract fine-grained features and ensure that densely arranged buildings can be accurately extracted and distinguished.By adding two different residual U-block(RSU),it increases the network′s sensitivity to small targets and uses an attention mechanism with different dimensions to pay reasonable attention to effective information.The experimental validation on the Massachusetts building dataset indicates the superiority of the proposed method.

关 键 词：智能巡检 建筑物提取 深度学习 语义分割 

分 类 号：TM755[电气工程—电力系统及自动化]